When a patient arrives in an emergency room with clinical symptoms consistent with bloodstream infection, blood cultures are drawn and empiric antimicrobial therapy is given;the actual identification of the pathogen by the laboratory typically takes one or more days. In the absence of specific data on the identity and susceptibility of the pathogen at the time of presentation, the clinician is forced to choose broad-spectrum antimicrobial therapy to cover all possible causes of the suspected bloodstream infection. Unfortunately, such empiric choices can sometimes end up being either ineffective (in the setting of antimicrobial resistance) or unnecessarily broad (in the setting of a susceptible and easily treated organism), potentially increasing morbidity, mortality, and resultant health care costs To address this need, we have developed a prototype identification system based on surface enhanced Raman spectroscopy (SERS). The detection technology consists of a portable Raman microscope, a novel nanostructured substrate, and detection algorithms that have exquisite analytical sensitivity (down to a single bacterium) and specificity (down to the strain level, with the ability to distinguish drug resistant bacteria). Moreover, detection is ultra-fast (~20 sec). To enable this technology to be used at point of care, we have developed an initial prototype system for isolating and concentrating low numbers of bacteria from blood and depositing those bacteria onto the SERS substrate within ~20 min. Here, we propose to build a next generation sample preparation prototype that will be integrated with our existing portable Raman microscope. Furthermore, we will increase the library of Raman signatures to include the most common causes of bacteremia and study the molecular basis for the signatures. The final system will be optimized and validated by testing with samples from human blood cultures and blood obtained directly from animals with experimental bacteremia. To meet these goals, we have assembled a multi-disciplinary team of engineers, basic scientists, and clinician-scientists. At the conclusion of this five-year project, we will have a hardened and tested system which will be ready for clinical studies to diagnose bacteremia in humans. The proposed system will enable identification of microbial pathogens rapidly enough to inform initial antimicrobial drug therapy, thereby reducing morbidity, mortality, and, thereby, healthcare costs. Moreover, the system can be used to address other types of infections by implementing minor changes to the sample processing system to handle additional sample types. We believe that the complete system will revolutionize the field of clinical microbiology by providing a new technology for identifying bacteria and providing basic susceptibility information in time for initial antimicrobial therapy. The system we propose to develop further with this grant will impact the diagnosis and treatment of bacterial infections by enabling clinicians to identify the infectious agent and its antibiotic resistance within half an hour. The physician can then use this information to correctly prescribe a narrow-spectrum antibiotic, which will result in improved patient care and reduced healthcare costs.